Human reasons to choose a human interpreter over machine interpretation

30 Jan 2018

Machine translation and interpreting systems have increased in both quantity and quality in recent years as the industry seeks solutions for the increasing demand of interpreting services worldwide. Google Translate is now commonly used for basic text translations over the internet, while voice recognition apps such as Lexifone and VoiceTra can provide machine-based interpretation.

The development of such technology has inevitably led to fears in some quarters about the future of human-based interpreting. Could increasingly refined machine interpretation eventually replace all human interpreting?

The reality is that this is very unlikely. Machines offer a fast and cheap alternative to human interpreters and will undoubtedly have a role to play in supplementing translation and interpreting services in the future. But tests have shown that humans heavily outperform machines when it comes to translation. For most professional interpreting needs, it’s difficult to imagine machines ever being capable of providing a service of sufficient accuracy or quality.  

Here are a few key reasons to choose human over machine interpretation.

Humans understand the differences between languages

Perhaps the first thing that should be pointed out is that translation and interpreting is not a science. It’s not simply a case of working out which words are said in one language and finding their exact equivalent in another. Each language has its own grammatical conventions, structural nuances, quirks and so on.

It’s difficult to get an exact translation using a machine as these nuances are missed. Machine technology aims for an exact literal translation but, because languages are different and exact literal translations aren’t always possible, what you end up with is a rough approximation of what is said. Occasionally, you’ll end up with something that makes no sense whatsoever. You can see this when using Google Translate or any machine interpretation system. The grammatical translation is rarely accurate and, if you use an unusual word or phrase, the system can break down completely.

Professional human interpreters, on the other hand, are fully trained in both languages. As they understand the nuances, conventions and idiomatic peculiarities of each language, they can offer a far better approximation of what’s being said. They can also tell when a literal translation is not possible and will be able to offer the nearest alternative instead.

Humans can interpret whereas machines can only translate

Language is more than just a set of words used for functional purposes. It’s a method of expression used to convey feelings and emotions that aren’t always readily picked up on a cold reading of the words being spoken. This is another advantage that humans have over machines. They are better able to understand the meaning behind the use of a word or phrase and interpret that meaning, whereas machines are not sophisticated enough for this.

This is especially important regarding words that can have multiple meanings. An understanding of the correct use of that word is crucial for the interpretation to be accurate. Words in languages such as English that have dual usage as nouns and verbs (such as ‘jam’, ‘bolt’ or ‘bark’) can lead to confusion with machine systems and produce inaccurate translations where the original meaning is totally lost.

Humans are able to grasp context

Another limitation with machine software is that it has no external awareness of anything beyond the language it is attempting to translate. Translating and interpreting duties are performed within a specific context which will undoubtedly have a bearing upon what is being said as well as how it is being said. Things to bear in mind include:

  • Local or cultural norms and practices which may influence use of language in terms of idioms or dialects.
  • The evolution of a language, with words sometimes changing meaning and new words and phrases being added. Human interpreters can more easily adapt to these changes.
  • Tone being used. For example, machines cannot detect jokes or sarcasm.
  • Who the audience is (age, education level, etc.)
  • The setting (is it formal such as a court or police office, or less formal such as an interview for a project or publication?)

Humans are able to differentiate between different settings, norms, etc. whereas machines are not.

Humans are more creative and expressive

There are many circumstances where interpreting becomes much more than simply offering up an approximation of what’s being said. It can be about capturing the heart and soul of something, whether it’s the story of someone who has endured something terrible or achieved something inspirational, or a particular brand identity of a company.

Human interpreters can gauge emotion and then look for the right words or expressions to best tell the story or portray the image. They can use appropriate metaphors or puns to bring things to life. The beauty of any story will most likely be lost through machine translation.

Humans can read non-verbal language

No matter how accurate or sophisticated machine interpreting technology becomes, one area it can’t pick up on (and is unlikely to be able to do so) is non-verbal language such as body language and facial expressions. It is estimated that around 55 percent of communication is non-verbal, rising to 93 percent if vocal elements such as tone are included. Much of this isn’t always spotted by humans, but translators and interpreters are more attuned to non-verbal signs and can do courses to raise their awareness.  

By picking up on conscious and subconscious non-verbal signs, human translators can enhance their understanding of communication and offer a level of service that goes beyond what machine interpretation is capable of.

Humans can engage with the subject

Human interpreters have the advantage of being able to engage with the subject in order to get clarity if something is misheard or misunderstood, or to help rectify an error. They can ask questions and build a dialogue to help improve the quality and accuracy of the interpretation. Just as written translators can review and edit their work to ensure the best quality translation, interpreters also need to review and amend at times in order to maintain standards. This is in contrast to machines where there is no process of refinement or of getting clarification as long as what has been said has been picked up by the voice recognition software.

Humans can show empathy

Because human interpreters can engage with those they are working with, they can also relate to them and show empathy. This can be very important in situations such as when interpreting for refugees or those who have suffered any form of abuse. In these circumstances, individuals may need a human face with whom they can build up some sort of trustful relationship rather than a machine before they can engage in dialogue. An understanding interpreter qualified to deal with such situations might also be able to provide additional forms of support or link their client up to other useful services.

The quality of machine synthesized speech is poor

When it comes to live interpreting, machines picking up and accurately interpreting what is being said is only half of the problem.  The systems also need to provide a verbal output. Speech synthesis has improved greatly in recent years but it is still a long way from acceptable standards in most professional fields in terms of pronunciation, tone and emphasis. Despite the efforts of developers, the best synthesized models have only been able to register sufficient levels of emotion in speech about 60 percent of the time.

Machines are not equipped to deal with human diversity

The diversity of the human race is such that there are over 6,000 languages in use on the planet today, with over 1,000 considered to be of significant global economic importance. For technology to be in a position to offer a viable alternative to human interpreting and translation, it would need to offer a service in all of these languages. Google Translate, which is the most commonly used machine translation service, currently operates in around 80 languages. This gives you some idea of how far machine interpreting has to go just to achieve global coverage. Bearing this in mind, the chances of machine technology ever catching up and replacing humans as interpreters seems very remote indeed.

You Might Also Like